Which is your pick? Both live on commonstack.ai now.

Opus 5. Refreshed homepage. More to come.

A glimpse into a month of production traffic through Commonstack: -12.9M+ requests -168B+ tokens -70 models -20%+ savings off list prices On to the next.

GPT-5.6 family is live on Commonstack. Route your agents to Sol, Terra and Luna. openai/gpt-5.6-sol openai/gpt-5.6-terra openai/gpt-5.6-luna Same key as your Claude, Gemini, DeepSeek and Kimi calls.

How do you evaluate an LLM router fairly? Most benchmarks look at prompts, but routers operate at an agentic-step level. A router that saves money but breaks the task could be worse than no router. We open-sourced TwinRouterBench to measure this honestly. 🧵

Fraction of the bill. Same results. Fully local, open source, works with any client. Just > pipx install uncommon-route github.com/CommonstackAI/Unc…

Run Claude Code with Commonstack in 4 steps: - generate an API key - set 4 environment variables - run claude - /status to verify Set it up now in 5 minutes with @alex_mirran.

Opus 4.7 shipped yesterday. It's now live on Commonstack, alongside 47 other models from 11 providers. Your frontier can be a full portfolio.

Model selection is driven by Thompson Sampling on real benchmark data (PinchBench), not price assumptions. Three feedback clicks change routing immediately. It gets smarter as you use it. Works with Codex, Claude Code, Cursor, OpenClaw, any OpenAI SDK client. pip install, set your upstream, serve.

Most routing tools match keywords. UncommonRoute has zero keyword lists. It's an ML classifier trained on structural features and character n-grams. No hardcoded rules. Works across 16 languages without language-specific config. Difficulty is a continuous 0-1 score, not a coarse label. Harder tasks automatically get more quality weight.

Not every prompt needs your most expensive model. ~80% of your requests are low-complexity steps: tool selection, result parsing, status checks. UncommonRoute routes those to cheaper models automatically to lower your API costs by >90% Open source, runs locally, MIT licensed.

If you're an OpenClaw🦞 builder and you've been hesitating about which model to pick, this quick comparison might help:

⚡ Attention OpenClaw builders 🦞 — GLM-5-Turbo works great with OpenClaw and can significantly accelerate agent workflows. @Zai_org It pairs naturally with OpenClaw and helps accelerate agent reasoning loops and tool usage. Try it on Commonstack.ai commonstack.ai/model-library…

Gemini 3 Flash ranked #1 on PinchBench on OpenClaw agent tasks success rate, followed by MiniMax M2.1 and Kimi K2.5. Don't forget Gemini 3 Flash Lite: same family, built for high-volume workloads at even lower prices. All on commonstack.ai

🦞Clawbox Updates🦞 > We’ve been seeing an explosion of new Skills across the OpenClaw ecosystem lately. You can now create and manage your own custom Skills directly in Clawbox > On the safety side, we also added support for customizable Soul.md with starter templates, so you can shape how your agent behaves from the start without needing to build everything from scratch.

Gemini 3.1 Pro is here @commonstack_ai Model highlights: > ARC-AGI-2 reasoning score jumps to 77.1% (~2× previous gen) > Wins 13/16 major benchmarks > API pricing unchanged Learn more about Commonstack.ai